Deep Reinforcement Learning from Policy-Dependent Human Feedback.
Dilip ArumugamJun Ki LeeSophie SaskinMichael L. LittmanPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- optimal policy
- markov decision process
- action selection
- policy search
- reinforcement learning algorithms
- reinforcement learning problems
- state and action spaces
- partially observable
- relevance feedback
- action space
- control policy
- reward function
- model free
- infinite horizon
- markov decision processes
- policy evaluation
- approximate dynamic programming
- partially observable domains
- multi agent
- control problems
- temporal difference
- human behavior
- state space
- human operators
- sensory inputs
- optimal control
- actor critic
- motor skills
- continuous state spaces
- human experts
- dynamic programming
- policy gradient methods